This job is writing the short sequence of emails a new customer receives in their first week, then setting your email tool to send it automatically. It is a Skill and Tech job that takes an afternoon: the AI drafts the sequence in one sitting, and the rest of the afternoon goes to loading it into your email tool and testing it. You give the AI the steps a new customer takes, the questions they ask most, a few emails you wrote yourself and the facts that must stay exact. It gives back three to five short emails, each with a subject line, a send day and one job to do. After that, the email tool sends the same first week to every new customer, and you come back to it when something in your business changes, or every few months to reread it.
- Shape
- SkillTech
- Sitting
- an afternoon
The shape: Skill and Tech
The Skill half is the writing. A first-week sequence is a small set of plain, practical emails with a known pattern: welcome, first step, common question, check-in. That is writing an AI does well when it has your real process in front of it. The skill you build here is the brief, and it carries over: when you add a new product or service, you reuse the same brief with different steps and get a second sequence in minutes.
The Tech half is the automation. The AI writes the emails, but it does not know when someone becomes a customer and it does not send anything. Your email tool does that, started by a trigger: a purchase, a booking, a form, or a tag your store or booking system adds. So the job ends inside Kit, Mailchimp, or the automated emails in your booking or billing platform, where you paste each email, set its delay and test the whole run.
What this means for you: the writing is the fast part. Budget most of the afternoon for the tool, because the trigger is where sequences quietly fail. If nothing in your setup can tell your email tool that a new customer exists, connecting that is a separate job that comes first.
What you give the AI
Gather these before you open the chat. The emails can only be as useful as the details you hand over.
- The steps a new customer takes after they say yes. Write them in order, in plain words: fill in the intake form, book the first visit, bring these two things, expect a call from us within two business days. If there are no steps, say what they should expect to happen and when.
- The three questions new customers ask most in their first week. Look at your inbox or your phone messages from recent new customers and copy the questions as they asked them. These questions are the strongest material you have.
- The one worry new customers have. Did they choose well, will this take a lot of their time, what happens if they need to cancel. Name it plainly so the sequence can answer it once, honestly.
- Two or three emails you wrote yourself. A reply to a new customer, a confirmation you typed by hand, a note to a regular. These teach the AI how you sound.
- Facts that must stay exact. Hours, phone number, the reply address, links to your booking page or help page, cancellation or refund terms, and the name of whoever answers. List them separately and mark them as facts not to change.
- Your email tool and its trigger. One line, such as "Kit, starts when the purchase tag is added" or "the booking platform's own automated emails." This tells the AI what the first email can assume.
- Anything off limits. A promise you will not make, a discount you do not want mentioned, a service you are phasing out.
Leave out anything private about real customers. Describe the questions, not the people who asked them.
What comes back
Paste this brief, then your inputs underneath it.
You are drafting the onboarding email sequence for a small business.
These emails go automatically to every new customer during their first week.
Write from the material below only. Do not add offers, numbers, steps, promises or policies that are not in the material.
Our business, in one line: [what you do and for whom]
The trigger that starts the sequence: [purchase, booking, form, tag]
Facts that must stay exactly as written: [hours, phone, links, terms, names]
Off limits: [anything not to mention]
Return 3 to 5 emails. For each email give:
1. Send timing, counted from the trigger (for example: immediately, day 1, day 3, day 6).
2. Its one job, in one sentence.
3. Two subject lines, under 50 characters, plain and specific.
4. One preview line, under 90 characters, that adds to the subject.
5. The email body, 60 to 180 words:
- Open with the reason for the email in the first sentence.
- One job per email and at most one link or one action.
- Short paragraphs. Write the way we write in the past emails below.
- No selling, no upsell, no "valued customer," no "journey."
The first email goes out immediately and confirms what happens next.
One email answers the most common question.
One email addresses the worry below, honestly, without overpromising.
The last email checks in and says how to reach a real person.
After the emails, return:
6. A list of every fact or promise in the sequence, with where it came from in our material.
7. Any question from our material you could not answer from the facts given.
The steps a new customer takes: [paste them in order]
The questions new customers ask: [paste them as asked]
The worry: [one line]
Our past emails, for voice: [paste two or three]What comes back is a short sequence you can read in five minutes, with a send day on each email and a list of every promise it makes. The list matters most, because each line points back to something you wrote. If a promise has no source in your material, the AI invented it, and it goes. The list of unanswered questions is useful too: it shows the gaps in your own process that a new customer would run into.
To move the sequence into your email tool, ask for each email again as plain text with every link written out. In the tool, create the automation, set the trigger, then add each email with the delay from its send timing. Set links inside the tool's editor rather than trusting pasted formatting. Then test the run end to end, which is the step most people skip.
How to judge it
A good sequence reads like a calm person walking a new customer through their first week. A bad one is easy to reject in a minute or two. Push back in the same chat, or start over with better inputs, when you see any of these.
- It promises something you do not do. A reply within the hour, a free extra, a guarantee you never offered. An automated email repeats the same wrong promise to every new customer, so this is the most serious failure.
- It sells in the first week. An upgrade, a referral ask or a second product before the customer has had the first one. Cut it, or move it to a later sequence.
- Every email says welcome. If the second and third emails open by welcoming the customer again, each one lacks a job. Ask for one job per email.
- It answers questions nobody asked. Generic tips about the industry instead of the questions you pasted. Point the AI back to your list and ask again.
- It sounds like nobody. "We are thrilled to have you on board," stacked adjectives, a closing line about the road ahead. Compare it with your own emails; if it does not sound like them, say so.
- It asks for three things at once. Fill in the form, follow us, leave a review and book a call in one email means none of them happens.
- It changed a fact you marked exact. A different phone number or a rounded refund window is a reason to reread every line of every email.
- The timing is crowded. Three emails in the first two days feels like a lot to someone who has just paid. Spread them out across the week.
If the sequence fails on voice, the fix is usually better inputs, not a better instruction. Give it a longer email you wrote by hand and ask again.
What stays with you
The AI drafts the emails. These parts stay yours, and no AI does them for you.
- The process itself. The AI cannot know what really happens after someone buys. If your first week is unclear to you, the sequence will be unclear to them, and the honest fix is to decide the steps first.
- Every promise and policy. Refund terms, response times, cancellation rules and anything a customer could hold you to get checked against what you actually do. An automated email makes the same promise to every new customer until you change it.
- The trigger and the test. Sign up as a test customer, or use the tool's preview with a test contact, with every delay set to a few minutes so the whole sequence arrives in one sitting, then set the real delays back. Click every link on your phone. Check that a customer who buys twice does not get the welcome twice.
- Consent and the list. An onboarding email goes to someone who has just bought from you, which is different from adding them to your marketing newsletter. Whether they also join the newsletter, and how they say yes to it, is your decision and your tool's settings, not the AI's.
- The replies. New customers reply to automated emails as if a person sent them, and a person should answer. Make sure the reply address goes to an inbox someone reads.
- The upkeep. When your hours, prices, links or steps change, the sequence keeps sending the old version until you edit it. Put a reminder on your calendar to reread it every few months.
An editorial example
This is an invented example, not a customer. Say you run a small dog grooming shop that sells a package of four grooms, booked online.
Your steps are: book the first appointment on the booking page, fill in the short form about the dog's coat and temperament, bring the vaccination record to the first visit. The questions new customers ask are whether they can stay during the groom, how long it takes, and what happens if the dog is nervous. The worry is that a nervous dog will have a bad first visit. Your booking platform adds a tag in your email tool when a package is bought, and that tag starts the sequence.
The first draft comes back with five emails, and the second one opens with "We are thrilled to welcome you to the family." You reject that and ask for one job per email. The second draft has four: an immediate email with the booking link and the form, a day-two email on what to bring and how long a groom takes, a day-four email about nervous dogs that describes what you actually do, and a day-six check-in with the shop's phone number and hours. The promise list shows that the nervous-dog email claimed "a free calming treatment," which is not in your material. You cut it.
You paste the four emails into your email tool, set the delays to a few minutes, run a test purchase with your own email address, and read all four on your phone that afternoon before you set the real delays. The booking link on the first email points to the wrong page, so you fix it before the first real customer arrives. That is the job.
Where the sequence runs
The Tech half is choosing where the sequence lives, and the best place is usually the tool that already knows when someone becomes a customer.
| If your new customers come through | Run the sequence in | Why |
|---|---|---|
| An online store or checkout that connects to Kit or Mailchimp | That email tool | The purchase can start the automation directly |
| A booking or scheduling platform with its own automated emails | The booking platform | It already knows the appointment date and can time emails around it |
| A form on your website | The email tool that receives the form | The form submission is the trigger |
| Phone calls and in-person sales only | One email tool, with customers added by hand | Adding each new customer starts the run; check the consent rules for your list first |
Plans and features change, so check each tool's own pages before you build on it.
Before you trust any sequence to run on its own, our page on email deliverability with AI covers checking that automated mail reaches the inbox. For the regular emails that go to everyone, writing your email newsletter with AI is the next job. When this one is done, the full list of DIY marketing jobs has the next one.
Of every job on this list, social media is the one that comes back every week, and Boomp is software that does that one for you. See ten posts made from your website, free
